Using Time Series Data and Machine Learning Estimating Agricultural Groundwater Extraction in Huwei Town, Taiwan

Groundwater is an essential source of water in Taiwan, and its long-term overuse has resulted in water resource problems that have become a potential crisis in the Zhuoshui River Basin. This overuse of groundwater may also lead to subsidence, which can have significant consequences for the area and its infrastructure. The lack of complete observations of groundwater extraction in Taiwan due to historical factors has made it difficult to accurately understand and manage the amount of water being taken, particularly for agricultural purposes.In view of this, this study uses time series data from 87 agricultural groundwater wells in Huwei Town, Yunlin County from January 2016 to July 2017, and time series data on agricultural well electricity usage in the Changshui River Basin, combined with other attribute data, to understand farmers' water pumping behavior using data mining methods and to estimate the amount of water taken in the Huwei area using machine learning.This study obtained the spatial and temporal distribution of groundwater withdrawals in the Huwei area in 2018.